Searcharxiv⌕ Search

arXiv subjects

M. Boiti

Publications and source records attributed to M. Boiti.

20 records · Page 2Linked to original sources

Properties Of Solutions Of The Kpi Equation

The Kadomtsev--Petviashvili I (KPI) is considered as a useful laboratory for experimenting new theoretical tools able to handle the specific features of integrable models in $2+1$ dimensions. The linearized version of the KPI equation is first considered by solving the initial value problem for different classes of initial data. Properties of the solutions in different cases are analyzed in details. The obtained results are used as a guideline for studying the properties of the solution $u(t,x,y)$ of the Kadomtsev--Petviashvili I (KPI) equation with given initial data $u(0,x,y)$ belonging to the Schwartz space. The spectral theory associated to KPI is studied in the space of the Fourier transform of the solutions. The variables $p=\{p_1,p_2\}$ of the Fourier space are shown to be the most convenient spectral variables to use for spectral data. Spectral data are shown to decay rapidly at large $p$ but to be discontinuous at $p=0$. Direct and inverse problems are solved with special attention to the behaviour of all the quantities involved in the neighborhood of $t=0$ and $p=0$. It is shown in particular that the solution $u(t,x,y)$ has a time derivative discontinuous at $t=0$ and that at any $t\not=0$ it does not belong to the Schwartz space no matter how small in norm and rapidly decaying at large distances the initial data are chosen.

patt-sol↗

Solutions of the Kpi Equation with Smooth Initial Data

The solution $u(t,x,y)$ of the Kadomtsev--Petviashvili I (KPI) equation with given initial data $u(0,x,y)$ belonging to the Schwartz space is considered. No additional special constraints, usually considered in literature, as $\int\!dx\,u(0,x,y)=0$ are required to be satisfied by the initial data. The problem is completely solved in the framework of the spectral transform theory and it is shown that $u(t,x,y)$ satisfies a special evolution version of the KPI equation and that, in general, $\partial_t u(t,x,y)$ has different left and right limits at the initial time $t=0$. The conditions of the type $\int\!dx\,u(t,x,y)=0$, $\int\!dx\,xu_y(t,x,y)=0$ and so on (first, second, etc. `constraints') are dynamically generated by the evolution equation for $t\not=0$. On the other side $\int\!dx\!\!\int\!dy\,u(t,x,y)$ with prescribed order of integrations is not necessarily equal to zero and gives a nontrivial integral of motion.

solv-int↗